Review of: Hieronymus F et al, Molecular Psychiatry 2017. doi:10.1038/mp.2017.147

Over the last several years, there’s been a lively debate about the efficacy of antidepressants. Meta-analyses have shown that antidepressants do outperform placebo in most studies. However, active medications cause more side effects than placebo pills. Thus, it’s possible that side effects lead patients to accurately guess that they’ve been assigned to the active drug group, and this positive expectancy alone could underlie any supposed efficacy of the drug.

But how would you structure a study to resolve this problem? You can’t randomly assign patients to the active medication with side effects vs the same active medication without side effects. So, researchers have resorted to the artful use of statistical methods. One of these is called a “mega-analysis.” In a mega-analysis, you combine the results of independent studies using data from the individual subjects. This differs from the more common meta-analysis, which combines the results of independent studies using summary data from groups of people.

In a new mega-analysis, the authors reviewed all industry-sponsored, FDA-registered, placebo-controlled trials of paroxetine and citalopram in adults with major depressive disorder. 2,759 patients were included from the paroxetine trials, and 585 patients formed the citalopram trials. The purpose was to compare the response of patients who reported side effects vs those who were side effect–free.

Focusing on the mood item of the Hamilton Depression Rating Scale (HDRS) as the outcome measure, researchers found that among patients who reported no side effects, patients assigned to paroxetine or citalopram did significantly better than those assigned to placebo (p < 0.001). Furthermore, looking at the entire sample, there were no significant differences in the reductions in the HDRS-17 depressed mood item at 6 weeks for paroxetine treatment with or without side effects, and for citalopram treatment with or without side effects.

TCPR’s Take
This study supports the hypothesis that both citalopram and paroxetine have genuine antidepressant effects caused by their pharmacodynamic properties, and more than just placebo effects.